TL;DR
Get everyday essentials delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
Amazon, Microsoft, Alphabet and Meta spent about $202 billion on property and equipment in July–September 2026, more than twice the year-earlier total, according to figures summed from company filings. They are financing the build-out with operating cash, debt, leases, partnerships and equity, while the scale and duration of future commitments leave questions about who bears the cost if AI revenue falls short.
Amazon, Microsoft, Alphabet and Meta spent about $202 billion on property and equipment in the third quarter of 2026, more than twice the roughly $97 billion they spent a year earlier, according to a sum of their cash-flow statements. The expansion, driven largely by data centers, servers and chips, is being financed through a mix of company cash, borrowing, leases, partnerships and share sales, spreading some of the cost beyond the technology firms themselves.
Several recent transactions show the shift. Meta sold $25 billion in common stock and $12.5 billion in preferred shares this week, while Microsoft issued $25 billion in bonds, including debt with a maturity extending to 2066. Amazon disclosed a loan facility of up to $80 billion secured by servers it plans to buy, according to the source article. These moves add financing options as spending rises, though their terms and purposes vary.
The companies’ third-quarter capital spending figures were $77.3 billion for Amazon, $51.4 billion for Alphabet, $39.4 billion for Microsoft and $33.9 billion for Meta. The totals are for property and equipment, not exclusively AI. Amazon also invests in warehouses; the source says the company described the recent increase as primarily reflecting AI investments. Amazon’s 2026 cash capital spending forecast is about $220 billion. Alphabet maintained its 2026 forecast of $195 billion to $205 billion and expects a significant increase in 2027.
Cash flow covers a different share of the bill at each company. In the quarter, Amazon and Alphabet spent more on property and equipment than their operating cash flow; Meta’s spending also slightly exceeded operating cash flow. Microsoft generated $58 billion in operating cash against $39.4 billion in spending. The source calculates Amazon’s trailing 12-month free cash flow at negative $42.4 billion, while Alphabet reported quarterly free cash flow of negative $9.1 billion and Meta negative $2.1 billion.
CAPITAL • CAPACITY • COMMITMENTS
Who Pays For AI’s Data Center Expansion? Trace The Money
Four U.S. tech giants spent about $202 billion on property and equipment in July–September 2026. Here’s how cash, debt, equity, leases and partners are financing the build-out—and where the long-term exposure may land.
01 / THE SPENDING SURGE
One quarter, four massive build-outs
Quarterly spending scale · $0–80B
Read the total carefully: these are property and equipment figures, not AI-only spending. Amazon also builds warehouses; it said the recent increase primarily reflected AI investments.
02 / FOLLOW THE FUNDING
Several channels are paying for capacity
Profits from advertising, software and commerce continue to fund infrastructure.
Microsoft issued $25B in bonds, including debt maturing in 2066.
Meta sold $25B in common stock and $12.5B in preferred shares.
Amazon disclosed a facility of up to $80B secured by servers it plans to buy.
Landlords, special-purpose companies and strategic partners can help finance facilities.
“Through a balanced mix of operating cash flow, investment-grade debt, strategic partnerships, and equity capital.”
Meta statement · October 26, 2026
03 / CASH FLOW & COMMITMENTS
Quarterly spending meets uneven cash coverage
In the quarter, Amazon and Alphabet spent more on property and equipment than their operating cash flow. Meta’s spending also slightly exceeded operating cash flow. Microsoft generated $58B in operating cash against $39.4B in spending.
The source reports Amazon’s trailing 12-month free cash flow at –$42.4B; Alphabet reported quarterly free cash flow of –$9.1B, and Meta –$2.1B.
Combined estimate across four quarterly filings as of September 30, mostly for data centers. It is future lease payments—not debt due immediately.
04 / THE DURATION MISMATCH
Some obligations may outlast the equipment
Reported or typical time horizons described in the source
Typical depreciation period
Long-lived supply commitments
Facilities may be leased long term
Microsoft bond maturity reaches 2066
If AI revenue grows more slowly than obligations, pressure could persist beyond the useful life of some servers. The actual allocation depends on contract terms, power markets and future demand.
05 / WHAT TO WATCH
Guidance offers near-term reference points
Amazon’s 2026 plan
About $220B in cash capital spending, according to its forecast.
Alphabet’s outlook
$195B–$205B for 2026, with a significant increase expected in 2027.
Higher-cost borrowing
GPU cloud provider CoreWeave’s senior notes carry 8.5%–9.75% interest, a different borrowing cost from the largest platforms.
The open question
Will AI revenue and customer demand keep pace with long-lived commitments for facilities and power?
What filings do not establish: the AI-only share of spending or leases, the revenue investments will produce, the net effect of facility income and customer contracts, or how much power infrastructure costs may reach electricity customers. Spending totals alone cannot settle whether the industry is in a bubble.
Trace the money—and the exposure
How the Financing Risk Spreads
The financing mix matters because the cost of building AI capacity extends beyond the current quarter’s cash spending. The companies also have future obligations for facilities and power. The source’s sum of four quarterly filings puts leases that have not yet started at about $1.02 trillion as of September 30, mostly for data centers. That figure is more than twice the companies’ combined long-term debt, though it represents future lease payments rather than debt due immediately.
Those commitments can involve different parties. Lenders buy bonds, investors provide equity, and landlords or special-purpose companies may own facilities leased to technology firms. Utilities and electricity customers may also be affected by the demand and infrastructure required to serve large data centers. The eventual distribution of costs depends on contract terms, power markets and whether companies can earn enough from AI services to support the build-out.
The time horizons are uneven: servers are typically depreciated over five to six years, while the source describes power contracts lasting 15 to 20 years, data center leases reaching 20 years and some bonds running 40 years. If AI revenue grows more slowly than the obligations, companies and their financing partners could face pressure over a longer period than the useful life of some equipment.
Popular right now
Dyson V8 Cyclone Cordless Vacuum
De'Longhi Magnifica Evo Fully Automatic Espresso Machine with Manual Milk Frother, Silver
Dyson Car+Boat Cordless Handheld Vacuum
As an affiliate, we earn on qualifying purchases.
From Cash-Funded Growth to Borrowing
The source describes a change from a period when large technology companies could pay for data centers from profits generated by advertising, software and online commerce. The 2026 spending surge has led them to use several funding channels at once. In a statement on October 26, Meta said it intended to finance AI investments through a “balanced mix of operating cash flow, investment-grade debt, strategic partnerships, and equity capital.”
Debt issuance is not limited to the largest platforms. The source identifies newer GPU cloud providers such as CoreWeave as another part of the financing picture and says its senior notes carry interest rates of 8.5% to 9.75%. Those rates indicate a different borrowing cost from that faced by the largest technology companies, but they do not alone establish how likely any borrower is to default.
The figures in the source are based on company filings and disclosures available in October 2026. The $202 billion spending total and $1.02 trillion lease total are the author’s sums across the four companies, rather than figures reported as a combined total by the companies themselves. Capital spending includes property and equipment that may serve uses beyond AI.
““Through a balanced mix of operating cash flow, investment-grade debt, strategic partnerships, and equity capital.””
— Meta, in a statement dated October 26
What the Commitments Do Not Show
The filings do not establish how much of the reported spending or lease commitments is solely for AI, or how much revenue the investments will ultimately generate. It is also unclear from the source how much of the future lease total will be offset by income from facilities, partnerships or customer contracts. The cited figures do not provide a combined estimate of power costs or identify which costs may ultimately reach electricity customers.
The scale of investment has prompted debate about whether the industry is in a bubble, but spending totals alone cannot settle that question. The financial exposure differs among technology firms, lenders, investors, landlords and infrastructure providers. The source identifies potential risk at those edges but does not quantify likely losses or show that any participant is currently unable to meet its obligations.
Company Guidance and Financing Ahead
Investors will be watching upcoming company filings and earnings updates for revised capital spending forecasts, operating cash flow and details of new borrowing or leasing arrangements. Amazon’s stated expectation of about $220 billion in cash capital spending for 2026 and Alphabet’s $195 billion to $205 billion range provide near-term reference points; Alphabet has also forecast a significant spending increase in 2027.
Further disclosures may clarify how much infrastructure is being built directly, leased or financed through partners, and how the companies expect AI-related income to cover those costs. The key open issue is whether revenue growth and customer demand will keep pace with long-lived commitments for facilities and power.
Key Questions
How much did the four companies spend in the third quarter?
Amazon, Microsoft, Alphabet and Meta spent a combined about $202 billion on property and equipment in July–September 2026, based on a sum of their cash-flow statements. The spending is not all necessarily for AI.
How are the companies paying for data centers?
The source identifies operating cash, debt, leases, strategic partnerships and equity as funding channels. Recent examples include Microsoft’s $25 billion bond sale, Meta’s stock and preferred share sales, and an Amazon loan facility of up to $80 billion secured by servers it plans to buy.
What does the $1.02 trillion lease figure mean?
It is the source author’s combined estimate of leases that had not yet started as of September 30, 2026, mostly for data centers. It represents future lease payments, not an immediate bill or a measure of AI revenue.
Who could bear the risk if AI revenue grows slowly?
Potentially the companies themselves, their lenders and investors, data center owners and other partners. Utilities and electricity customers could also be affected by infrastructure and power demand, but the source does not quantify those costs.
Source: Thorsten Meyer AI
